Loading...
Loading...
Transcriptomic datasets of human primary blood–derived macrophages were identified through the GEO repository. These included macrophages polarized in vitro to M1 (stimulated with lipopolysaccharide, interferon-γ, or both) or M2 (stimulated with IL-4, IL-13, or both) phenotypes. Eight RNA-seq and thirteen microarray datasets were selected, totaling 57 M0 (unstimulated), 92 M1, and 87 M2 samples.
Gene expression data were downloaded, merged into a unified matrix, and filtered to exclude low-informative features such as microRNAs, long non-coding RNAs, ribosomal proteins, and hypothetical transcripts. Samples were annotated based on phenotype (M0, M1, M2), treatment condition, and GEO dataset identifier. Genes not detected in at least 20 samples per condition were excluded. Batch effects across studies were corrected using the `removeBatchEffect` function from the limma package.
Differential gene expression between groups (M1 vs. M0, M2 vs. M0, and M1 vs. M2) was assessed using empirical Bayes statistics implemented in the limma package. P-values were adjusted for multiple testing using the Bonferroni method.
Genes were considered specific to the M1 or M2 polarization states if they met the following criteria: FDR < 10⁻³ and log₂(fold change) > 0 compared to M0, and FDR < 10⁻³ and log₂(fold change) > 2 when comparing M1 and M2. The resulting gene signature lists were used to test for enrichment in external RNA sequencing datasets using Fisher’s exact test.
Download the human signature data and statistics Download the mouse signature data and statisticsPillon NJ, Smith JAB, Alm PS, Chibalin AV, Alhusen J, Arner E, Carninci P, Fritz T, Otten J, Olsson T, van Doorslaer de Ten Ryen S, Deldicque L, Caidahl K, Wallberg-Henriksson H, Krook A, Zierath JR. Distinctive exercise-induced inflammatory response and exerkine induction in skeletal muscle of people with type 2 diabetes. Sci Adv. 2022 Sep 9;8(36):eabo3192.